A chemically synthesized DNA template, poly(dA), and calf thymus DNA were purchased from Sigma-Aldrich Inc. and a customized oligo(dT)₁₈ DNA primer was produced by Sigma-Aldrich Japan K.K. (Hokkaido, Japan). Radioactive nucleotide [³H]-labeled 2′-deoxythymidine-5′-triphosphate (dTTP; 43 Ci/mmol) was obtained from Moravek Biochemicals Inc. (Brea, CA, USA). Supercoiled pHOT-1 DNA was obtained from TopoGEN Inc. (Port Orange, FL, USA). All other reagents were analytical grade from Nacalai Tesque Inc. (Kyoto, Japan).

Some xanthone compounds were isolated from the leaves of mangosteen (Garcinia mangostana L.), which were collected in Vietnam and provided by Dr Duy Hoang Le (Kobe Pharmaceutical University). The chemical structures were determined by spectroscopic analyses and comparison with reported data (27–30). The identified eight compounds are α-mangostin (1), β-mangostin (2), 3,6-di-O-methyl-γ-mangostin (3), fuscax-anthone C (4), gartanin (5), 8-deoxygartanin (6), mangostanin (7) and morusignin I (8) (Fig. 1).

Pol α was purified from calf thymus by immunoaffinity column chromatography, as described by Tamai et al (31). Recombinant rat Pol β was purified from Escherichia coli JMpβ5, as described by Date et al (32). The human Pol γ catalytic gene was cloned into pFastBac, with the histidine-tagged enzyme expressed using the BAC-TO-BAC HT Baculovirus Expression System according to the supplier's instructions (Life Technologies, Inc., Gaithersburg, MD, USA), and purified using ProBound resin (Invitrogen Japan K.K, Tokyo, Japan) (33). Human Pols δ and ɛ were purified by nuclear fractionation of human peripheral blood cancer cells (Molt-4) using the second subunit of Pols δ and ɛ-conjugated affinity column chromatography, respectively (34). A truncated form of human Pol η (residues 1–511) tagged with His₆ at the C-terminus was expressed in E. coli and purified as described by Kusumoto et al (35). Recombinant mouse Pol ι tagged with His₆ at the C-terminus was expressed and purified by Ni-NTA column chromatography (36). A truncated form of Pol κ (residues 1–560) with a His₆-tag at the C-terminus was expressed in E. coli and purified as described by Ohashi et al (37). Recombinant human His-Pol λ was expressed and purified according to a method described by Shimazaki et al (38). Calf TdT, T7 RNA polymerase and T4 polynucleotide kinase were purchased from Takara Bio Inc. (Kyoto, Japan). Purified human placenta Topos I and II were purchased from TopoGEN Inc. Bovine pancreas deoxyribonuclease I was obtained from Stratagene Cloning Systems (La Jolla, CA, USA).

The reaction mixtures for calf Pol α and rat Pol β have been described (39,40). Those for human Pol γ and for human Pols δ and ɛ were as described by Umeda et al (33) and Ogawa et al (41), respectively. The reaction mixtures for mammalian Pols η, ι and κ were the same as that for calf Pol α, and the reaction mixture for human Pol λ was the same as for rat Pol β. For Pols, poly(dA)/oligo(dT)₁₈ (A/T=2/1) and tritium-labeled 2′-deoxythymidine 5′-triphos-phate ([³H]-dTTP) were used as the DNA template-primer and nucleotide [i.e., 2′-deoxynucleoside 5′-triphosphate (dNTP)] substrates, respectively. For the TdT reactions, oligo(dT)₁₈ (3′-OH) and dTTP were used as the DNA primer substrate and nucleotide substrate, respectively.

The eight isolated xanthone compounds were dissolved in distilled dimethyl sulfoxide (DMSO) at various concentrations and sonicated for 30 sec. Aliquots of 4 μl of sonicated samples were mixed with 16 μl of each enzyme (final amount, 0.05 units) in 50 mM Tris-HCl (pH 7.5), containing 1 mM dithiothreitol, 50% glycerol and 0.1 mM ethylenediaminetetraacetic acid (EDTA), and kept at 0ºC for 10 min. These inhibitor-enzyme mixtures (8 μl) were added to 16 μl of each of the enzyme standard reaction mixtures and incubation carried out at 37ºC for 60 min. Activity without the inhibitor was considered 100% and the remaining activity at each inhibitor concentration was determined relative to this value. One unit of Pol activity was defined as the amount of enzyme that catalyzed incorporation of 1 nmol dNTP [measured using [³H]-dTTP] into synthetic DNA template-primers in 60 min at 37ºC under the normal reaction conditions for each enzyme (39,40).

Purified human placental Topos I and II were purchased from TopoGen Inc. The catalytic activity of Topo I was determined by detecting supercoiled plasmid DNA (Form I) in its nicked form (Form II) (42). The Topo I reaction was performed in a 20-μl reaction mixture that contained 10 mM Tris-HCl (pH 7.9), pHOT-1 DNA (250 ng), 1 mM EDTA, 150 mM NaCl, 0.1% bovine serum albumin (BSA), 0.l mM spermidine, 5% glycerol, 2 μl of one of the eight isolated xanthone compounds dissolved in DMSO, and 2 units of Topo I. Topo II catalytic activity was analyzed in the same manner, except the reaction mixture contained 50 mM Tris-HCl (pH 8.0), 120 mM KCl, 10 mM MgCl₂, 0.5 mM ATP, 0.5 mM dithiothreitol, supercoiled pHOT-1 DNA (250 ng), and 2 units of Topo II (42). Reaction mixtures were incubated at 37ºC for 30 min, followed by digestion with 1% sodium dodecyl sulfate (SDS) and 1 mg/ml proteinase K. After digestion, 2 μl loading buffer (5% sarkosyl, 0.0025% bromophenol blue and 25% glycerol) was added. The same procedure was followed for mobility shift assay assessment of enzyme DNA binding, except SDS denaturation and proteinase K digestion were omitted. Reaction mixtures were subjected to 1% agarose gel electrophoresis in Tris/borate/EDTA buffer. Agarose gels were stained with ethidium bromide (EtBr) and DNA band shifts from Form I to Form II were detected using an enhanced chemiluminescence detection system (Perkin-Elmer Life Sciences Inc., Waltham, MA, USA). Zero-D scan (Version 1.0, M & S Instruments Trading Inc., Osaka, Japan) was used for densitometric quantitation.

The activities of T7 RNA polymerase, T4 polynucleotide kinase and bovine deoxyribonuclease I were measured using standard assays according to the manufacturer's specifications as described by Nakayama and Saneyoshi (43), Soltis and Uhlenbeck (44) and Lu and Sakaguchi (45), respectively.

Thermal transition profiles of dsDNA to single-stranded DNA (ssDNA) with or without α-mangostin were obtained with a spectrophotometer (UV-2500; Shimadzu Corp., Kyoto, Japan) equipped with a thermoelectric cell holder was used as previously described (46). Calf thymus DNA (6 μg/ml) was dissolved in 0.1 M sodium phosphate buffer (pH 7.0) containing 1% DMSO. The solution temperature was equilibrated at 75ºC for 10 min, and then increased by 1ºC at 2-min intervals for each measurement point. Any change in the absorbance (260 nm) of the test compound itself at each temperature point was automatically subtracted from that of DNA plus the compound in the spectrophotometer.

The human cervical cancer cell line HeLa was obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin. Cells were cultured at 37ºC in a humidified incubator containing 5% CO₂/95% air. For assessment of cell growth, cells were plated at 2×10³ cells/well in 96-well microplates, cultured for 12 h and various concentrations of the eight isolated xanthone compounds added. The compounds were dissolved in DMSO to produce 10-mM stock solutions, which were further diluted to appropriate final concentrations with growth medium containing 0.5% DMSO immediately prior to use. Cell viability was determined by the WST-1 assay following 24-h incubation (47).

Cellular DNA content for cell cycle analysis was determined as follows: aliquots of 3×10⁵ HeLa cells were added to a 35-mm dish and incubated with the medium that contained the test compound for 0 to 24 h. Cells were then washed with ice-cold PBS three times, fixed with 70% (v/v) ethanol, and stored at −20ºC. DNA was stained with a PI (3,8-diamino-5-[3-(diethylmethylammonio)propyl]-6-phenyl-phenanthridinium diiodide) staining solution for at least 10 min at room temperature in the dark. Fluorescence intensity was measured using a BD FACSCalibur flow cytometer in combination with CellQuest software (Becton-Dickinson Co., Tokyo, Japan).

The enzymatic activity of caspase-3 was measured using EnzChek Caspase-3 Assay Kit #2 (Life Technologies Japan Co., Ltd., Tokyo, Japan), according to the instruction manual. HeLa cells (1×10⁶) after inducing apoptosis with test compound were washed with PBS and lysed. Then, the caspase-3 activity in the extracts was determined with a fluorometric assay. The fluorescent product of the substrate Z-DEVD-Rhodamine 110 generated by caspase-3 in the cell extract was detected in a fluorescent microplate reader (SH-9000 Lab; Hitachi High-Technologies Corp., Tokyo, Japan) using an excitation of 496 nm and emission of 520 nm. To determine the background fluorescence of the substrate, test compound without the Z-DEVD-R110 substrate for negative controls, and mixture of activity buffer and Z-DEVD-R110 substrate for only substrate controls were also executed.

Aliquots of 2.5×10⁴ HeLa cells were plated in each well of an 8-well chamber slide (Thermo Fisher Scientific K.K., Tokyo, Japan). Then, the cells were incubated with or without the test compound (27.2 and 54.4 μM) for 12 and 24 h at 37ºC with 5% CO₂. After washed with PBS twice, the cells were fixed with 4% paraformaldehyde (pH 7.4) for 15 min. Then, the cells were stained with DAPI for 1 min, and the percentage of apoptotic cells was counted under a fluorescence microscope (Olympus IX70; Olympus, Tokyo, Japan).

The data are expressed as the mean value ± the standard deviation (SD) of the mean of at least three independent determinations for each experiment. Statistical significance between each experimental group was analyzed using Student's t-test, and a probability level of 0.01 and 0.05 was used as the criterion of significance.

Selective inhibitors of mammalian Pols have been investigated with the goal to develop chemothera-peutic drugs for anticancer, antivirus and anti-inflammation treatments (48). In our ongoing screening for natural inhibitors of mammalian Pols, we found that the extract of mangosteen (Garcinia mangostana L.) had inhibitory activity against mammalian Pols. The isolated xanthone compounds 1-8 (Fig. 1) were selected and prepared for this study.

The inhibitory activity of each compound toward mammalian Pols was investigated using calf Pol α, rat Pol β, and human Pols γ and κ. For mammalian Pols, Pols α, β, γ and κ were used as the representatives of Pol families B, X, A and Y, respectively (6,7). Assessment of the relative activity of each Pol at a set concentration (10 μM) of the eight test compounds showed that β-mangostin (2) was the strongest inhibitor of these Pol species among the compounds tested (Fig. 2). In particular, this compound showed the strongest inhibitory activity against Pol α among the four mammalian Pols. α-Mangostin (1) and 3,6-di-O-methyl-γ-mangostin (3) slightly inhibited Pols α, γ, and κ activities and Pol κ activity, respectively. These results suggested that the common backbone structure with prenyl groups at C-2 and C-8 of compounds 1-3 might be important for Pol inhibition, and both the hydroxyl group at C-6 and the methoxy group at C-3, which are present in β-mangostin (2), must be much more important for Pol inhibitory activity.

Next, the inhibitory effects of the eight isolated xanthones (10 μM each) were examined against human Topos I and II, which have ssDNA and dsDNA nicking activity, respectively (8). β-Mangostin (2) was the strongest inhibitor of both Topos I and II, and the inhibitory effect of this compound on Topo II was more potent than that on Topo I (Fig. 3). In contrast, mangostanin (7) and morusignin I (8) did not influence the activities of these Topos. Topo inhibition could be ranked as β-mangostin (2) >α-mangostin (1) >3,6-di-O-methyl-γ-mangostin (3) = fuscaxanthone C (4) = gartanin (5) = 8-deoxygartanin (6) >mangostanin (7) = morusignin I (8). Since β-mangostin (2) showed the strongest inhibitory activities on both mammalian Pols and human Topos, we focused on this compound in the subsequent study.

Evaluation of the inhibition of DNA metabolic enzyme activities in vitro by β-mangostin revealed that this compound inhibited the activities of all eleven mammalian Pols tested from all four families. Calf Pol α and calf TdT were the strongest and weakest inhibited, respectively, with 50% inhibitory concentration (IC₅₀) values of 6.4 and 39.6 μM, respectively (Table I). The inhibition by β-mangostin against mammalian Pol families could be ranked as B-family Pols >A-family Pol >Y-family Pols > X-family Pols. By comparison, the IC₅₀ values of aphidicolin, a known eukaryotic DNA replicative Pols α, δ and ɛ inhibitor, were 20, 13 and 16 μM, respectively (49), showing that the Pol inhibitory activity of β-mangostin was >1.5-fold higher than that of aphidicolin. β-Mangostin inhibited the activity of human Topos I and II with IC₅₀ values of 10.0 and 8.5 μM, respectively. Topotecan and doxorubicin, which are Topo I and Topo II inhibitors, respectively, also inhibited the nicking activities of Topos I and II with IC₅₀ values of 45 and 60 μM, respectively (data not shown). The data show that the inhibitory effect on Topos I and II of β-mangostin was more potent than that of topotecan and doxorubicin, respectively. The inhibition of human Topos activities by β-mangostin was approximately the same order of potency compared with that of the activities of mammalian Pols except for X-family of Pols (Table I).

In contrast, β-mangostin did not influence the activities of other DNA metabolic enzymes, such as T7 RNA polymerase, T4 polynucleotide kinase, and bovine deoxyribonuclease I (Table I). These results indicate that β-mangostin can be specifically classified as an inhibitor of mammalian Pols and Topos.

Specific assays were performed to determine whether β-mangostin-induced inhibition resulted from the ability of the compound to bind to DNA or the enzyme. The interaction of β-mangostin with dsDNA was investigated by studying its thermal transition. The melting temperature (Tm) of dsDNA in the presence of an excess of β-mangostin (200 μM) was observed using a spectrophotometer equipped with a thermoelectric cell holder. As shown in Fig. 4, a thermal transition (i.e., Tm) from 75 to 90ºC was not observed within the concentration range used in the assay, whereas when a typical intercalating compound, such as EtBr (15 μM), was used as a positive control, an obvious thermal transition was observed.

We then assessed whether the inhibitory effect of β-mangostin on mammalian Pols and Topos resulted from non-specific adhesion to the enzyme or from selective binding to specific sites. Neither excessive nucleic acid [in the form of Poly(rC)] nor protein (BSA) significantly influenced β-mangostin-induced both Pol and Topo inhibition (data not shown). This suggests that β-mangostin selectively bound to the Pol and Topo molecules. These observations reveal that β-mangostin does not act as a DNA intercalating agent, and rather it directly binds the enzyme to inhibit its activity.

Collectively, these results suggested that β-mangostin might be a potent and selective inhibitor of mammalian Pols and Topos. We next investigated whether Pol and/or Topo inhibition by β-mangostin resulted in decreases in human cancer cell proliferation.

Pols and Topos have recently emerged as important cellular targets for the development of anticancer agents. Therefore, we also investigated whether the eight isolated xanthones had any cytotoxic effect on HeLa cells. At 100 μM, all the compounds except for 3,6-di-O-methyl-γ-mangostin (3) potently suppressed the rate of HeLa cell growth to <80% that of control (Fig. 5). At 10 μM, β-mangostin (2) could moderately suppress HeLa cell growth, but other compounds did not influence cell proliferation. The 50% lethal dose (LD₅₀) of β-mangostin (2) was 27.2 μM, which is ~2-fold higher than the IC₅₀ for DNA replicative Pols (X-family Pols α, δ and ɛ) and Topos. This suggests that β-mangostin may be able to penetrate the cell membrane and reach the nucleus, where it may then inhibit Pols and/or Topos activities to suppress cell growth. Similar findings for all eight xanthones were observed in the HCT116 human colon carcinoma cell line (data not shown). These results reveal that Pol and Topo inhibition (Figs. 2 and 3) and human cancer cell cytotoxicity (Fig. 5) can be induced by xanthones from mangosteen. Therefore, Pol/Topo inhibition by β-mangostin represents a potentially useful anticancer effect.

Next we analyzed whether β-mangostin could affect the cell cycle distribution of HeLa cells. The cell cycle fraction was recorded after 0, 6, 12 and 24 h of treatment with a concentration of β-mangostin at 54.4 μM (the double density of LD₅₀). The ratio of cells in each of the three cell cycle phases (G1, S and G2/M) is shown in Fig. 6A. Treatment with β-mangostin significantly increased the population of cells in S and G1 phase at 12 and 24 h, respectively, but did not influence the rate of G2/M phase. Dehydroaltenusin, which is a mammalian Pol α specific inhibitor, arrested the cell cycle in the G1 and S phases (50), and etoposide, which is a classical Topo II inhibitor, arrested the cell cycle in the G2/M phase (data not shown). These results suggest that β-mangostin may be an effective inhibitor of DNA replicative Pols, such as Pol α, that arrests the cell cycle at the G1 and S phases.

To examine whether the cell cycle arrest of HeLa cells treated with β-mangostin was due to apoptosis, caspase-3 activity was analyzed using fluorescent substrate (Fig. 6B). When HeLa cells were treated with the compound at 27.2 (=LD₅₀ value) and 54.4 μM, the cells maximally increased caspase-3 activity for 24 and 12 h, respectively (Fig. 6B). In addition, preincubation with a caspase-3 inhibitor, DEVD-CHO, completely blocked this activity by β-mangostin (data not shown), suggesting that the apoptosis by this compound was mediated through caspase-3 activation.

Fig. 6C and D shows the extent of apoptotic cells by DAPI staining. The percentage of HeLa apoptotic cells after treatment with 27.2 and 54.4 μM of β-mangostin was markedly high for 24 and 12 h, respectively (Fig. 6C). These findings suggest that the human cancer cytotoxic effect of β-mangostin must involve a combination of cell proliferation arrest and apoptotic cell death.